Hi,
I'm refining a structure that contains heavy atoms as well as BH4 anions. I try to preserve the expected geometry of the BH4 units by heavily weighted restraints on the B-H distances and H-B-H angles with the macros Distance_Restrain and Angle_Restrain. All the restraints are initially satisfied, i.e. they give zero penalties.
When I allow refinement of B- and H-atom positions (or only B-atom positions) for several iterations, the restraints are violated and at the same time Rwp
increases slightly. chi and P (sum of penalties) get enormous due to the heavy weighting of the restraints. This link shows a screenshot of the output:
https://imgur.com/a/za20d7K
It seems to me that things get out of hand in the very first iteration. In iteration zero, Rwp = 8.04 which is then also the value of chi since the penalty sum is zero. Rwp falls to 7.35 in the first iteration. At the same time, the penalty sum gets sky-high (P 3.9E8). But despite this, chi remains equal to Rwp! So the least-square refinement doesn't seem to care about the restrains at this point. In the next iteration, chi catches up and gets very high, but the refinement never recovers and ends up with both higher Rwp and massive restraints violations.
Is it a bug or a feature that the restraints are not influencing chi in the first iteration? And what is the best way to deal with it? Rigid body description would of course keep things in place, but I find it a bit awkward that I can't force atoms to stay in special positions in the final stage of the refinement.